A digital subscriber line (DSL) technology is a broadband access technology that implements high-speed data service transmission by using a common phone twisted pair (unshielded twist pair, UTP) as a transmission medium. The digital subscriber line technology mainly enables, by using a frequency division multiplexing technology, a digital subscriber line service and a plain old telephone service (POTS) to coexist in a same common phone twisted pair without a need of replacing an existing basic transmission medium; and in addition, when data transmission is performed by using an existing common phone twisted cable, a high-speed uplink/downlink transmission rate can be achieved. In a whole transmission band, the POTS service occupies a baseband part less than 4 KHz, while the digital subscriber line service occupies a band of a high-frequency part; and signals from the POTS service and the digital subscriber line service are separated by using a separator. The digital subscriber line technology mainly uses a discrete multi-tone (DMT) modulation scheme, so as to improve an anti-interference capability of a digital subscriber line system.
Referring to FIG. 1A and FIG. 1B, a figure in FIG. 1A is a schematic diagram of near-end crosstalk and a figure in FIG. 1B is a schematic diagram of far-end crosstalk. In a digital subscriber line system, a digital subscriber line access multiplexer (DSLAM) 101, as a central office device of the digital subscriber line system, can access multiple DSL lines and optimize a transmission rate. Two DSL lines, namely, a first line 103 established between a central office transceiver 1011 and a subscriber end 1021 and a second line 104 established between a central office transceiver 1012 and a subscriber end 1022, are used as an example. According to an electromagnetic induction principle, crosstalk interference is generated between two signals of the first line 103 and the second line 104 that are accessed by the DSLAM. The crosstalk interference is classified into far-end crosstalk, (FEXT) and near-end crosstalk (NEXT), where FEXT refers to interference between uplink signals of different wire pairs or between downlink signals of different wire pairs, and NEXT refers to interference between an uplink signal and a downlink signal of different wire pairs.
FEXT and NEXT are both enhanced as a frequency band increases; however, because an uplink/a downlink channel of a digital subscriber line system uses a frequency division multiplexing manner, NEXT can be eliminated or reduced by using a filter, which does not affect the digital subscriber line system greatly. However, as a frequency band used by the digital subscriber line system is increasingly larger, FEXT is also enhanced gradually. It can be learn from the Shannon equation C=B·log 2 (1+S/N) (where C is a channel rate, B is a signal bandwidth, S is signal energy, and N is noise energy) that, a greater N indicates a smaller C; and during digital subscriber line transmission, crosstalk interference acts as a noise part, and stronger FEXT indicates a greater N; therefore, severe FEXT may remarkably reduce the channel rate. In this way, when multiple subscribers in a bundle of cables request to enable a digital subscriber line service, due to FEXT, transmission rates of some lines are low, performance is unstable, and even the service cannot be enabled, thereby finally resulting in a low service-enabling rate of a DSLAM.
In view of the foregoing problems, the industry proposes a Vectoring technology currently, in which a DSLAM end uses a downlink precoding technology and an uplink joint reception technology to implement crosstalk cancellation, crosstalk vector information in a line is acquired by means of interaction between the DSLAM end and a terminal, and then an “inverse” crosstalk signal is acquired by performing complex matrix calculation and the “inverse” crosstalk signal is then superimposed on a digital subscriber line signal. During transmission of the digital subscriber line signal, the “inverse” crosstalk signal and FEXT in the line cancel each other out, thereby reducing impact of FEXT on line transmission performance.
In a downlink direction, a precoding technology is used to perform crosstalk cancellation on a to-be-sent signal, so that a signal received by a receive end is not affected by crosstalk interference from another line. However, power and power spectrum density (PSD) at which signals are sent by a transmit end in the downlink direction are strictly limited, and total power for sending the signals cannot exceed a specified maximum value (for example, in 17a template of VDSL2, a power maximum value is 14.5 dBm, that is,
                              10                      14.5            10                          ⁢                                  ⁢        mW            ≈              28.1838        ⁢                                  ⁢        mW              )    ,which requires that precoded downlink signals do not increase total transmit power. To solve a problem of a power limitation, by using linear precoding processing performed on a signal as an example, a precoded signal is generally amplified or diminished in a normalized manner by using a normalization factor λ, so that a sent signal of each line can meet a limitation requirement of the PSD, that is, the total transmit power does not exceed the specified maximum value; and a received signal is recovered by using a recovery factor
  1  λat a receive end, so as to avoid signal distortion. When the normalization factor λ changes, the recovery factor
  1  λat the receive end also correspondingly changes. Ideally, the transmit end and the receive end respectively use the normalization factor λ and the recovery
  1  λfactor at the same time, which requires that the normalization factor λ used by the transmit end and the recovery factor
  1  λused by the receive end are necessarily synchronized and coordinated to ensure that distortion, or even a bit error, of a received signal does not occur at the receive end. In addition, when the recovery factor
  1  λat the receive end changes, a signal-to-noise ratio (SNR) of a line also changes, which probably needs to change a related parameter currently used by the receive end; otherwise, distortion of the received signal is probably caused. In this way, it is required that the transmit end and the receive end can synchronously adjust the related parameter to avoid a bit error of the signal received by the receive end. Therefore, due to the foregoing reasons, a mechanism for jointly and synchronously adjusting parameters of a transmit end and a receive end of a line is required.